- Email: [email protected]
Avoiding futile damage control laparotomy
Injury, Int. J. Care Injured 41 (2010) 64–68
Contents lists available at ScienceDirect
Injury journal homepage: www.elsevier.com/locate/injury
Avoiding futile damage control laparotomy Nicolas Kairinos *, Philip M. Hayes, Andrew J. Nicol, Delawir Kahn Trauma Centre, Dept of Surgery, Groote Schuur Hospital and University of Cape Town, Cape Town, South Africa
A R T I C L E I N F O
A B S T R A C T
Article history: Accepted 26 May 2009
Background: The age of a patient, lowest pre-operative pH and lowest core temperature are significant predictors of mortality in patients undergoing damage control surgery (DCS). An equation had previously been devised based on these three variables, which could predict which patients would die despite undergoing DCS (100% positive predictive value, 25% sensitivity). The aim of this study was to validate this equation by testing it on a different cohort of patients undergoing DCS. Patients and methods: A retrospective validation study of patients who underwent DCS over a four-year period (1998–2001) was undertaken. The lowest pre-operative pH, lowest pre-operative core temperature and age were recorded and the equation was used to predict which patients were ‘‘unsalvageable’’. This was then compared to the true outcomes of these patients. Results: A total of 73 case notes were analysed for the period 1998–2001. The equation predicted that eight patients were unsalvageable. All eight patients died (100% positive predictive value), despite DCS being utilised. A further 25 of the rest of the ‘‘potentially salvageable’’ patients also died (24% sensitivity). When data of the original study (2002–2004) was combined with the current study data, the cohort totalled 145 patients, of whom 53 died (37%). Thirteen of these would have been predicted as unsalvageable with a 100% positive predictive value, had the equation been used during this time. Conclusion: Both the positive predictive value and sensitivity of the equation remain consistent. When resources are overwhelmed by multiple casualties, this equation could prove useful in identifying patients in whom surgery may be futile, allowing surgical triage to be directed in a more efficient manner. ß 2009 Elsevier Ltd. All rights reserved.
Keywords: Damage control surgery Damage control laparotomy Abbreviated laparotomy Abbreviated celiotomy Trauma laparotomy Predicting mortality
Introduction Although the concept of the abbreviated laparotomy was introduced decades earlier, the term ‘‘damage control surgery’’ (DCS) was popularised by Rotondo et al.15 The aim of DCS is avoiding the ‘‘triad of death’’, namely hypothermia, acidosis and coagulopathy, and this has proven to be life-saving in patients who would otherwise have had mortality rates as high as 90%.7 This is achieved by limiting the initial surgery to the control of haemorrhage and contamination from bowel content. Only once the patient’s metabolic disturbances have been rectified in an intensive care unit, is the patient returned to theatre for a re-look, usually at about 36–48 h. Despite improved survival rates following the advent of DCS, a collective review published in 2000 quoted mortality rates
* Corresponding author at: Dept of Surgery, J-Floor, Old Main Building, Groote Schuur Hospital, Main Rd, Observatory, 7925, Cape Town, South Africa. Tel.: +27 21 4043426; fax: +27 21 552 7310. E-mail addresses: [email protected], [email protected] (N. Kairinos). 0020–1383/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.injury.2009.05.036
averaging about 50%.17 These patients may survive their initial injuries but there appears to be a two-hit phenomenon, whereby many of these survivors, later die as a result of multi-organ failure (MOF) and sepsis. Blood and blood products, emergency list theatre time, nursing staff and ICU beds are some of the scarce resources on which this form of surgery is most taxing.16 They may require in excess of 40 units of blood within the first hour of surgery4 and a total of 10 operations.11 A single patient undergoing DCS may cost in excess of £50 000 depending their length of admission and pathology. Some may spend up to 90 days in ICU and then eventually die due to persistent complications related to their initial surgery.11 For these reasons, there have been a number of studies that have speculated on the possibility of predicting the outcome of these patients.1,4 When analysing predictive factors, there appears to be individual factors that have ‘‘cut off’’ values, beyond which, mortality rates approach 100%, despite DCS being performed. Temperature appears to be one of these, and in trauma surgery, and particularly DCS, 32 8C appears to be a critical value, with mortality being inevitable at lower temperatures.10 Studies have also shown that increasing age is independently a significant predictor of mortality,11,13 with some suggesting that performing DCS on
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
patients older than 70 years is futile.14 Similarly, as pH drops, mortality increases despite DCS being performed.1,3–5,11 There may be a cut off value for this too, whereby a lower pH represents such advanced cellular decompensation beyond which no human could survive. Each one of these variables has been shown to be a powerful, independent predictor of mortality; therefore, if a patient were to have a poor combination of all three of these variables it could be conceived that their survival would be even less likely. It is on this premise that the damage control equation is based. In a previous study at this institution, age, worst pre-operative pH and lowest pre-operative core temperature were all shown to be significant, independent predictors of ultimate mortality, despite survival of the initial damage control operation.11 As all three of these variables are known prior to surgery, it was postulated that it may be possible to preoperatively predict which patients would not survive this course of management. A mathematical model was devised, based on these variables, which accurately predicted ultimate mortality in the group of patients that had the greatest degree of physiological decompensation. When used on that data, it had a 100% positive predictive value of mortality. The aim of this study was to validate the devised equation by testing it on a different cohort of patients, in order to assess whether it retained a 100% positive predictive value. The study was approved by the Institutional Review Board. Patients and methods Development of the equation In the previous retrospective study on patients who underwent DCS at this centre (n = 72), more than 30 demographic and physiological variables and their correlation with mortality were analyzed. Of the variables that could be determined preoperatively, it was found that age (p < 0.0001), lowest preoperative core temperature (p = 0.001) and worst pre-operative pH (p < 0.0001) were independently significant predictors of mortality.11 A linear discriminant analysis was applied to these data. The derived equation described a discriminant function that uses the variables that had the best discrimination between those who died and those who survived, i.e. the best combination of sensitivity (at identifying futile cases) and positive predictive value for mortality. By adjusting the constant in the equation, this equation could be biased in a direction allowing an increase in the positive predictive value (at the cost of decreasing the sensitivity). Using the cohort of damage control patients from the previous study, it was found that when the constant was set at 6.002 the positive predictive value of the equation was 100% (Fig. 1). The sensitivity of predicting death was 25%, implying that of all the patients that die, only the quarter with the worst combination of age, temperature and pH would be predicted to die. The derived equation can be seen in Fig. 2. Study methodology Complete case notes of all patients undergoing damage control surgery from January 1998 to December 2001 were analysed retrospectively. Age, lowest pre-operative pH and lowest preoperative core temperature values were determined. These values and the ultimate outcome of the patient were then logged into a database by the second author. The first author, blinded to the true outcome of the patients, applied the equation to these variables, attempting to predict whether a patient would have been potentially salvageable or unsalvageable. The predictions were then compared to the patients’ true outcomes. Although numerous factors, e.g. patient demographics, mechanism of injury, vital signs,
65
Fig. 1. A scatter plot indicating that when the predictive equation’s constant is set at 6002, all patients who scored values above 0.5 died. In the group scoring less than or equal to 0.5 there were survivors and non-survivors.
Fig. 2. The derived damage control equation. X > 0.5 indicates inevitable mortality. Therefore, by rounding off X to the nearest whole number, X = 0 implies the patient is potentially salvageable, while X = 1 implies certain mortality.
blood results, trauma scores, etc., were analysed on the cohort of patients from the initial study (2002–2004), these were not taken into account when analysing the second cohort. The reason for this was that the concept of the DCS equation was that it should be possible to apply it to any patient who is so moribund that DCS is required, regardless of other factors. Although only the current study’s data (1998–2001) was used for validation purposes, both studies’ data were later combined to illustrate the total number of patients that would have been positively predicted as unsalvageable over a seven-year period (1998–2004). Results Complete case notes were obtained for 73 patients undergoing DCS during January 1998 and December 2001. The equation predicted that eight of these were unsalvageable, whilst the rest were potentially salvageable. All of the eight predicted unsalvageable patients died, despite DCS being utilised (100% positive predictive value). An additional 25 patients belonging to the potentially salvageable group also died, i.e. a sensitivity of 24%. By adding this data to the initial cohort of patients (2002–2004, n = 72) from the previous study, the combined cohort consisted of 145 DCS patients (1998–2004). There were 53 deaths, of which 13 were predicted by the equation, thus a 25% sensitivity and 100% positive predictive value (Fig. 3). Ten of the 13 died within the first 24 h. One died on the fifth post-operative day and the other two on day 15. Discussion Having predicted eight of the 33 deaths in the current study, the equation’s sensitivity (24%) appears to have changed very little from that of the original study (25%). The fact that all eight of the predictions were correct demonstrates that the equation consistently retains the 100% positive predictive value, even when used on a different cohort of patients undergoing DCS. Had this equation
66
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
Fig. 3. Proportion of non-survivors predicted by equation.
been utilised in its current form between 1998 and 2004, it would have predicted the deaths of 13 out of 145 patients. Surgeons are skilled at knowing how and when to operate but mastering the skill of knowing when not to operate is often difficult. In trauma, this is even more difficult and consequently, practically all critically traumatised patients are operated on. The concept of clinical decisions being based on a mathematical equation requiring only three variables may appear rash on initial consideration. However, appreciation of the nature of the pathology involved and careful consideration of the presented research, suggests that this concept of preoperative prediction may well be plausible. Increasing age has been found to be significantly predictive of outcome following trauma11,13 and therefore forms part of most trauma scoring systems. The decrease in physiological reserve appears to be the most logical reason for this. The effect that age has on the outcome of DCS, as opposed to other trauma surgery, is likely to be more pronounced. Age was independently the most significant preoperative predictor of mortality in research done at this unit.11 Over the seven-year period, no patient older than 58 years of age survived DCS (Fig. 4). With regards to core temperature, it is the lowest value that appears to be most predictive of mortality,2,3,5,9–11,13 implying that there seems to be a degree of irreversible physiological damage that occurs once this variable deteriorates beyond a certain point. Jurkovich et al., studied the relationship between hypothermia and survival in trauma patients and observed that mortality was 100% in patients with a core body temperature less than 32 8C.10 Our study echoed these findings, with 32 8C appearing to be a critical value (Fig. 5). It has been suggested that the level of hypothermia observed is the result of cellular decompensation rather than the cause of this.12 It seems that, if physiological decompensation in trauma victims were to occur to such a degree that cellular
Fig. 4. Relationship of age and mortality in patients undergoing damage control surgery (n = 145).
Fig. 5. Relationship of temperature (8C) and mortality in patients undergoing damage control surgery (n = 145).
metabolism could not generate sufficient energy to maintain a core temperature of more than 32 8C, then mortality rates approach 100%. The cellular decompensation results in hypothermia, which, in turn, further compromises the precarious physiological state of the patient by affecting enzymatic functions, clotting cascades, platelet function, cardiac output, etc. The deterioration of these systems deprives cells of their requirements even further, exacerbating the extent of cellular decompensation. It can therefore be seen how temperature homeostasis can rapidly spiral out of control, with hypothermia becoming refractory to even the most aggressive rewarming techniques. Gregory et al., found that the mean temperature loss was greater in the emergency department than in the operating room.9 This was also our experience, implying that the lowest preoperative temperature appears to be the most crucial of the temperature measurements. Acid base status has also been implicated as an important predictor of mortality by numerous researchers.1,3–5,11 Like hypothermia, acidosis also adversely affects enzymatic functioning, clotting, myocardial contractility etc. In the previous study from this institution, it was found that pH was a greater independent predictor of mortality than base excess.11 This may be due to the fact that values such as base excess or lactate are more indicative of volume status. pH is influenced not only by volume status but by respiratory status too. A patient that has a severe volume deficit is likely to be more compromised if there is concomitant respiratory embarrassment from chest trauma or cardiac failure. This may explain why pH was found to have greater statistical significance as a predictor of mortality than base excess in the previous study.11 Although others have suggested that pH can be confounded in patients with head injuries due to iatrogenic mandatory hyperventilation,8 this did not appear to be a factor in this study, perhaps because the majority of our patients suffered penetrating injuries (59/72) with only eight percent (6/72) requiring preoperative intubation. In a study by Burch et al.,4 it appears that a pH of less than 6.8 is not compatible with life, despite DCS. In our study too, none of the patients survived if the pH was less than 6.85 (Fig. 6). In a study by Asensio et al. (n = 548),3 one patient was found to survive surgery with a pH of 6.65 and another with a temperature of 30.5 8C. However, it was not indicated whether these patients underwent DCS. Furthermore, not all of the three parameters being considered in this study (age, pH and temperature) were mentioned for each of these two patients. Those parameters may have been in favour of the patient’s survival in these cases. This emphasises the importance of having all three of these important variables in one equation. Although there are many trauma scores in existence, all are used primarily for audit and research purposes. Clinical decisions are seldom, if ever, based on these scores. The current damage control equation is unique in a number of aspects. Unlike other scoring systems, which can offer a prognosis in the form a percentage, this equation identifies a small subgroup of patients
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
Fig. 6. Relationship of pH and mortality in patients undergoing damage control surgery (n = 145).
that have been demonstrated to be unsalvageable, despite the fact that DCS had been performed. Furthermore, it is the only scoring system that can be calculated preoperatively. This allows for decision making prior to committing to a very costly course of management. An additional advantage of this equation is its simplicity. The calculation can be undertaken within minutes if necessary, which, in the trauma resuscitation setting can be very useful. However, there are significant ethical issues. It may be argued that there is the potential for incorrect management decisions if the equation is based solely on single values. Because of possible human or technical errors in taking the measurement, it would be advocated that if a patient were predicted to be unsalvageable, a repeat temperature and pH ought to be taken immediately to confirm the initial findings. It must be stressed that this is for confirmation only and not to observe a trend. In the original study, it was observed that patients who were predicted as unsalvageable based on their worst values, died, even if these values improved following resuscitation. Therefore, observing a trend over time appears to have little merit. A further argument against the concept of this equation is that it is possible that not all patients fit within the normogram that this equation is based on and that a false positive prediction of death might occur. This may become apparent if a larger study population were to be evaluated, preferably as part of a multicentre study. This, however, did not appear to be the case in the combined total of 145 patients at this centre. Furthermore, it must be borne in mind that 30% (40/132) of the group that were predicted as salvageable, died. The equation’s low sensitivity (25%) to predict mortality is desirable, as this further reduces the possibility of incorrect prediction of death. There are some very thought-provoking ethical arguments, however, that can be made in defence of such an equation. In this study, the positively predicted patients had a 100% mortality rate, despite having had DCS. In resource-limited situations, this would raise the question of whether the surgery in this subgroup of patients was futile or not. To assist in answering this question, consideration needs to be given to what there is to lose in initiating this management. Firstly, the numerous surgical and other interventions that these patients undergo, and determination of the surgical team, may give the family false hope that the patient may survive. The surgeon may be merely drawing out the inevitable and, in some instances, subjecting the patient to an undignified death. Although this is unlikely to have been an issue in most of the positively predicted patients (10/13 died within 24 h), there were a few that survived for days or even weeks. DCS also places a strain on resources such as blood products, ICU beds, theatre time, available surgeons and nursing requirements.4,16 The use of DCS on an unsalvageable patient may therefore be to the detriment of other patients, particularly in the
67
multiple-casualty setting.6 In this setting, the natural tendency of surgeons is to operate on the sickest patient first, which may be an unsalvageable patient. During this time a critically ill, but salvageable, patient may become unsalvageable if there was a delay in the damage control procedure taking place due to limited resources. In most first world settings, with abundant resources, such an equation may find little benefit. However, under-resourced hospitals commonly seen in the third world could benefit from such an objective tool, particularly when overwhelmed by extreme situations, such as natural disasters or acts of terrorism. Although this equation is an exciting prospect in the field of trauma surgery it remains only a concept at this stage and there are certain issues that need to be considered. It is based on the principle that, at some point, the patient’s physiological deterioration seems to become refractory to any form of management. However, advances in DCS techniques and post-operative resuscitation could conceivably help salvage such a patient in the future. Should such advances occur, the equation may need to be revised. Some may recommend that the equation be used retrospectively, to guide further post-operative management. The patient may, for example, undergo the initial DCS operation despite a positive prediction. If, however, the patient continues to deteriorate in the intensive care unit, the predicted outcome of the equation may help further decision making. The coefficients in this particular equation were derived from data from one centre. It may be that, with different levels of care in more advanced centres and in different population groups, the survival rates are different. The equation’s coefficients would need to be adjusted to suit that group’s particular data. It should be noted, however, that if a particular equation is derived from data from an advanced unit, which is experienced in treating DCS patients, then the potential for false positive predictions is likely to be even lower if the same equation were to be used at a less advanced unit, thereby making it safer to use. This implies that such an equation should ideally be derived from data from advanced units, preferably multiple centres, in so doing making it unlikely to give a false positive prediction in less advanced units. It is recommended that the current DCS equation described in this paper be tested in a larger, multicentre, prospective trial. This particular equation was derived in a centre with a high incidence of penetrating trauma (82%).11 Although there were no false positive predictions in patients suffering blunt trauma, testing larger cohorts of patients in this blunt trauma group would be preferable. Even if this particular equation undergoes radical change, the principle it is based on still remains, and is one that should not be discarded. That is, all human beings have a limit to what they can endure in the trauma setting, albeit with a wide variation between individuals. A point must exist however, beyond which no human being can survive their physiological decompensation and, until now we have had no way of knowing this point. The current damage control equation appears to be a step in that direction and is at the very least a preliminary definition of that point. Conclusion Despite this being one of the largest predictive studies in damage control surgery, further work and larger numbers from multiple centres are certainly required. These preliminary data, however, suggest that using the damage control equation may offer a means of predicting mortality preoperatively with a high degree of certainty. In this study, the equation predicted a quarter of all patients that died without any incorrect predictions. Although such an equation is unlikely to be used under normal circumstances, it may be useful in crisis situations when services are overwhelmed by large numbers of severely injured casualties.
68
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
Conflict of interest None of the authors have any conflict of interest due to financial or personal relationships with other people or organisations that could inappropriately bias their work. There were no study sponsors. Acknowledgement Dr. Sedick Isaacs, for reviewing the statistics and assistance in formulating the equation. References 1. Aoki N, Wall MJ, Demsar J, et al. Predictive model for survival at the conclusion of a damage control laparotomy. Am J Surg 2000;180(December (6)):540–4 [discussion 544–5]. 2. Arthurs Z, Cuadrado D, Beekley A, et al. The impact of hypothermia on trauma care at the 31st combat support hospital. Am J Surg 2006;191(May (5)):610–4. 3. Asensio JA, McDuffie L, Petrone P, et al. Reliable variables in the exsanguinated patient which indicate damage control and predict outcome. Am J Surg 2001;182(December (6)):743–51. 4. Burch JM, Ortiz VB, Richardson RJ, et al. Abbreviated laparotomy and planned reoperation for critically injured patients. Ann Surg 1992;215(May (5)):476–83 [discussion 483–4]. 5. Cosgriff N, Moore EE, Sauaia A, et al. Predicting life-threatening coagulopathy in the massively transfused trauma patient: hypothermia and acidoses revisited. J Trauma 1997;42(May (5)):857–61 [discussion 861–2].
6. Eiseman B, Moore EE, Meldrum DR, Raeburn C. Feasibility of damage control surgery in the management of military combat casualties. Arch Surg 2000;135(November (11)):1323–7. 7. US Department of Defence. Emergency war surgery. 3rd ed. Virtual Naval Hospital. www.vhh.org; 2007. 8. Falcone RE, Santanello SA, Schulz MA, et al. Correlation of metabolic acidosis with outcome following injury and its value as a scoring tool. World J Surg 1993;17(September–October (5)):575–9. 9. Gregory JS, Flancbaum L, Townsend MC, et al. Incidence and timing of hypothermia in trauma patients undergoing operations. J Trauma 1991;31(June (6)):795–8 [discussion 798–800]. 10. Jurkovich GJ, Greiser WB, Luterman A, Curreri PW. Hypothermia in trauma victims: an ominous predictor of survival. J Trauma 1987;27(September (9)):1019–24. 11. Kairinos N, Nicol AJ, Timmermans J, Navsaria PH. Predicting mortality in damage control surgery. J Surg Res 2007;137(2):240. 2. 12. Krishna G, Sleigh JW, Rahman H. Physiological predictors of death in exsanguinating trauma patients undergoing conventional trauma surgery. Aust N Z J Surg 1998;68(December (12)):826–9. 13. Luna GK, Maier RV, Pavlin EG, et al. Incidence and effect of hypothermia in seriously injured patients. J Trauma 1987;27(September (9)):1014–8. 14. Morris Jr JA, Eddy VA, Blinman TA, et al. The staged celiotomy for trauma. Issues in unpacking and reconstruction. Ann Surg 1993;217(May (5)):576–84 [discussion 584–6]. 15. Rotondo MF, Schwab CW, McGonigal MD, et al. ‘Damage control’: an approach for improved survival in exsanguinating penetrating abdominal injury. J Trauma 1993;35(September (3)):375–82 [discussion 382–3]. 16. Sagraves SG, Toschlog EA, Rotondo MF. Damage control surgery—the intensivist’s role. J Intensive Care Med 2006;21(January–February (1)):5–16. 17. Shapiro MB, Jenkins DH, Schwab CW, Rotondo MF. Damage control: collective review. J Trauma 2000;49(November (5)):969–78.
Contents lists available at ScienceDirect
Injury journal homepage: www.elsevier.com/locate/injury
Avoiding futile damage control laparotomy Nicolas Kairinos *, Philip M. Hayes, Andrew J. Nicol, Delawir Kahn Trauma Centre, Dept of Surgery, Groote Schuur Hospital and University of Cape Town, Cape Town, South Africa
A R T I C L E I N F O
A B S T R A C T
Article history: Accepted 26 May 2009
Background: The age of a patient, lowest pre-operative pH and lowest core temperature are significant predictors of mortality in patients undergoing damage control surgery (DCS). An equation had previously been devised based on these three variables, which could predict which patients would die despite undergoing DCS (100% positive predictive value, 25% sensitivity). The aim of this study was to validate this equation by testing it on a different cohort of patients undergoing DCS. Patients and methods: A retrospective validation study of patients who underwent DCS over a four-year period (1998–2001) was undertaken. The lowest pre-operative pH, lowest pre-operative core temperature and age were recorded and the equation was used to predict which patients were ‘‘unsalvageable’’. This was then compared to the true outcomes of these patients. Results: A total of 73 case notes were analysed for the period 1998–2001. The equation predicted that eight patients were unsalvageable. All eight patients died (100% positive predictive value), despite DCS being utilised. A further 25 of the rest of the ‘‘potentially salvageable’’ patients also died (24% sensitivity). When data of the original study (2002–2004) was combined with the current study data, the cohort totalled 145 patients, of whom 53 died (37%). Thirteen of these would have been predicted as unsalvageable with a 100% positive predictive value, had the equation been used during this time. Conclusion: Both the positive predictive value and sensitivity of the equation remain consistent. When resources are overwhelmed by multiple casualties, this equation could prove useful in identifying patients in whom surgery may be futile, allowing surgical triage to be directed in a more efficient manner. ß 2009 Elsevier Ltd. All rights reserved.
Keywords: Damage control surgery Damage control laparotomy Abbreviated laparotomy Abbreviated celiotomy Trauma laparotomy Predicting mortality
Introduction Although the concept of the abbreviated laparotomy was introduced decades earlier, the term ‘‘damage control surgery’’ (DCS) was popularised by Rotondo et al.15 The aim of DCS is avoiding the ‘‘triad of death’’, namely hypothermia, acidosis and coagulopathy, and this has proven to be life-saving in patients who would otherwise have had mortality rates as high as 90%.7 This is achieved by limiting the initial surgery to the control of haemorrhage and contamination from bowel content. Only once the patient’s metabolic disturbances have been rectified in an intensive care unit, is the patient returned to theatre for a re-look, usually at about 36–48 h. Despite improved survival rates following the advent of DCS, a collective review published in 2000 quoted mortality rates
* Corresponding author at: Dept of Surgery, J-Floor, Old Main Building, Groote Schuur Hospital, Main Rd, Observatory, 7925, Cape Town, South Africa. Tel.: +27 21 4043426; fax: +27 21 552 7310. E-mail addresses: [email protected], [email protected] (N. Kairinos). 0020–1383/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.injury.2009.05.036
averaging about 50%.17 These patients may survive their initial injuries but there appears to be a two-hit phenomenon, whereby many of these survivors, later die as a result of multi-organ failure (MOF) and sepsis. Blood and blood products, emergency list theatre time, nursing staff and ICU beds are some of the scarce resources on which this form of surgery is most taxing.16 They may require in excess of 40 units of blood within the first hour of surgery4 and a total of 10 operations.11 A single patient undergoing DCS may cost in excess of £50 000 depending their length of admission and pathology. Some may spend up to 90 days in ICU and then eventually die due to persistent complications related to their initial surgery.11 For these reasons, there have been a number of studies that have speculated on the possibility of predicting the outcome of these patients.1,4 When analysing predictive factors, there appears to be individual factors that have ‘‘cut off’’ values, beyond which, mortality rates approach 100%, despite DCS being performed. Temperature appears to be one of these, and in trauma surgery, and particularly DCS, 32 8C appears to be a critical value, with mortality being inevitable at lower temperatures.10 Studies have also shown that increasing age is independently a significant predictor of mortality,11,13 with some suggesting that performing DCS on
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
patients older than 70 years is futile.14 Similarly, as pH drops, mortality increases despite DCS being performed.1,3–5,11 There may be a cut off value for this too, whereby a lower pH represents such advanced cellular decompensation beyond which no human could survive. Each one of these variables has been shown to be a powerful, independent predictor of mortality; therefore, if a patient were to have a poor combination of all three of these variables it could be conceived that their survival would be even less likely. It is on this premise that the damage control equation is based. In a previous study at this institution, age, worst pre-operative pH and lowest pre-operative core temperature were all shown to be significant, independent predictors of ultimate mortality, despite survival of the initial damage control operation.11 As all three of these variables are known prior to surgery, it was postulated that it may be possible to preoperatively predict which patients would not survive this course of management. A mathematical model was devised, based on these variables, which accurately predicted ultimate mortality in the group of patients that had the greatest degree of physiological decompensation. When used on that data, it had a 100% positive predictive value of mortality. The aim of this study was to validate the devised equation by testing it on a different cohort of patients, in order to assess whether it retained a 100% positive predictive value. The study was approved by the Institutional Review Board. Patients and methods Development of the equation In the previous retrospective study on patients who underwent DCS at this centre (n = 72), more than 30 demographic and physiological variables and their correlation with mortality were analyzed. Of the variables that could be determined preoperatively, it was found that age (p < 0.0001), lowest preoperative core temperature (p = 0.001) and worst pre-operative pH (p < 0.0001) were independently significant predictors of mortality.11 A linear discriminant analysis was applied to these data. The derived equation described a discriminant function that uses the variables that had the best discrimination between those who died and those who survived, i.e. the best combination of sensitivity (at identifying futile cases) and positive predictive value for mortality. By adjusting the constant in the equation, this equation could be biased in a direction allowing an increase in the positive predictive value (at the cost of decreasing the sensitivity). Using the cohort of damage control patients from the previous study, it was found that when the constant was set at 6.002 the positive predictive value of the equation was 100% (Fig. 1). The sensitivity of predicting death was 25%, implying that of all the patients that die, only the quarter with the worst combination of age, temperature and pH would be predicted to die. The derived equation can be seen in Fig. 2. Study methodology Complete case notes of all patients undergoing damage control surgery from January 1998 to December 2001 were analysed retrospectively. Age, lowest pre-operative pH and lowest preoperative core temperature values were determined. These values and the ultimate outcome of the patient were then logged into a database by the second author. The first author, blinded to the true outcome of the patients, applied the equation to these variables, attempting to predict whether a patient would have been potentially salvageable or unsalvageable. The predictions were then compared to the patients’ true outcomes. Although numerous factors, e.g. patient demographics, mechanism of injury, vital signs,
65
Fig. 1. A scatter plot indicating that when the predictive equation’s constant is set at 6002, all patients who scored values above 0.5 died. In the group scoring less than or equal to 0.5 there were survivors and non-survivors.
Fig. 2. The derived damage control equation. X > 0.5 indicates inevitable mortality. Therefore, by rounding off X to the nearest whole number, X = 0 implies the patient is potentially salvageable, while X = 1 implies certain mortality.
blood results, trauma scores, etc., were analysed on the cohort of patients from the initial study (2002–2004), these were not taken into account when analysing the second cohort. The reason for this was that the concept of the DCS equation was that it should be possible to apply it to any patient who is so moribund that DCS is required, regardless of other factors. Although only the current study’s data (1998–2001) was used for validation purposes, both studies’ data were later combined to illustrate the total number of patients that would have been positively predicted as unsalvageable over a seven-year period (1998–2004). Results Complete case notes were obtained for 73 patients undergoing DCS during January 1998 and December 2001. The equation predicted that eight of these were unsalvageable, whilst the rest were potentially salvageable. All of the eight predicted unsalvageable patients died, despite DCS being utilised (100% positive predictive value). An additional 25 patients belonging to the potentially salvageable group also died, i.e. a sensitivity of 24%. By adding this data to the initial cohort of patients (2002–2004, n = 72) from the previous study, the combined cohort consisted of 145 DCS patients (1998–2004). There were 53 deaths, of which 13 were predicted by the equation, thus a 25% sensitivity and 100% positive predictive value (Fig. 3). Ten of the 13 died within the first 24 h. One died on the fifth post-operative day and the other two on day 15. Discussion Having predicted eight of the 33 deaths in the current study, the equation’s sensitivity (24%) appears to have changed very little from that of the original study (25%). The fact that all eight of the predictions were correct demonstrates that the equation consistently retains the 100% positive predictive value, even when used on a different cohort of patients undergoing DCS. Had this equation
66
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
Fig. 3. Proportion of non-survivors predicted by equation.
been utilised in its current form between 1998 and 2004, it would have predicted the deaths of 13 out of 145 patients. Surgeons are skilled at knowing how and when to operate but mastering the skill of knowing when not to operate is often difficult. In trauma, this is even more difficult and consequently, practically all critically traumatised patients are operated on. The concept of clinical decisions being based on a mathematical equation requiring only three variables may appear rash on initial consideration. However, appreciation of the nature of the pathology involved and careful consideration of the presented research, suggests that this concept of preoperative prediction may well be plausible. Increasing age has been found to be significantly predictive of outcome following trauma11,13 and therefore forms part of most trauma scoring systems. The decrease in physiological reserve appears to be the most logical reason for this. The effect that age has on the outcome of DCS, as opposed to other trauma surgery, is likely to be more pronounced. Age was independently the most significant preoperative predictor of mortality in research done at this unit.11 Over the seven-year period, no patient older than 58 years of age survived DCS (Fig. 4). With regards to core temperature, it is the lowest value that appears to be most predictive of mortality,2,3,5,9–11,13 implying that there seems to be a degree of irreversible physiological damage that occurs once this variable deteriorates beyond a certain point. Jurkovich et al., studied the relationship between hypothermia and survival in trauma patients and observed that mortality was 100% in patients with a core body temperature less than 32 8C.10 Our study echoed these findings, with 32 8C appearing to be a critical value (Fig. 5). It has been suggested that the level of hypothermia observed is the result of cellular decompensation rather than the cause of this.12 It seems that, if physiological decompensation in trauma victims were to occur to such a degree that cellular
Fig. 4. Relationship of age and mortality in patients undergoing damage control surgery (n = 145).
Fig. 5. Relationship of temperature (8C) and mortality in patients undergoing damage control surgery (n = 145).
metabolism could not generate sufficient energy to maintain a core temperature of more than 32 8C, then mortality rates approach 100%. The cellular decompensation results in hypothermia, which, in turn, further compromises the precarious physiological state of the patient by affecting enzymatic functions, clotting cascades, platelet function, cardiac output, etc. The deterioration of these systems deprives cells of their requirements even further, exacerbating the extent of cellular decompensation. It can therefore be seen how temperature homeostasis can rapidly spiral out of control, with hypothermia becoming refractory to even the most aggressive rewarming techniques. Gregory et al., found that the mean temperature loss was greater in the emergency department than in the operating room.9 This was also our experience, implying that the lowest preoperative temperature appears to be the most crucial of the temperature measurements. Acid base status has also been implicated as an important predictor of mortality by numerous researchers.1,3–5,11 Like hypothermia, acidosis also adversely affects enzymatic functioning, clotting, myocardial contractility etc. In the previous study from this institution, it was found that pH was a greater independent predictor of mortality than base excess.11 This may be due to the fact that values such as base excess or lactate are more indicative of volume status. pH is influenced not only by volume status but by respiratory status too. A patient that has a severe volume deficit is likely to be more compromised if there is concomitant respiratory embarrassment from chest trauma or cardiac failure. This may explain why pH was found to have greater statistical significance as a predictor of mortality than base excess in the previous study.11 Although others have suggested that pH can be confounded in patients with head injuries due to iatrogenic mandatory hyperventilation,8 this did not appear to be a factor in this study, perhaps because the majority of our patients suffered penetrating injuries (59/72) with only eight percent (6/72) requiring preoperative intubation. In a study by Burch et al.,4 it appears that a pH of less than 6.8 is not compatible with life, despite DCS. In our study too, none of the patients survived if the pH was less than 6.85 (Fig. 6). In a study by Asensio et al. (n = 548),3 one patient was found to survive surgery with a pH of 6.65 and another with a temperature of 30.5 8C. However, it was not indicated whether these patients underwent DCS. Furthermore, not all of the three parameters being considered in this study (age, pH and temperature) were mentioned for each of these two patients. Those parameters may have been in favour of the patient’s survival in these cases. This emphasises the importance of having all three of these important variables in one equation. Although there are many trauma scores in existence, all are used primarily for audit and research purposes. Clinical decisions are seldom, if ever, based on these scores. The current damage control equation is unique in a number of aspects. Unlike other scoring systems, which can offer a prognosis in the form a percentage, this equation identifies a small subgroup of patients
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
Fig. 6. Relationship of pH and mortality in patients undergoing damage control surgery (n = 145).
that have been demonstrated to be unsalvageable, despite the fact that DCS had been performed. Furthermore, it is the only scoring system that can be calculated preoperatively. This allows for decision making prior to committing to a very costly course of management. An additional advantage of this equation is its simplicity. The calculation can be undertaken within minutes if necessary, which, in the trauma resuscitation setting can be very useful. However, there are significant ethical issues. It may be argued that there is the potential for incorrect management decisions if the equation is based solely on single values. Because of possible human or technical errors in taking the measurement, it would be advocated that if a patient were predicted to be unsalvageable, a repeat temperature and pH ought to be taken immediately to confirm the initial findings. It must be stressed that this is for confirmation only and not to observe a trend. In the original study, it was observed that patients who were predicted as unsalvageable based on their worst values, died, even if these values improved following resuscitation. Therefore, observing a trend over time appears to have little merit. A further argument against the concept of this equation is that it is possible that not all patients fit within the normogram that this equation is based on and that a false positive prediction of death might occur. This may become apparent if a larger study population were to be evaluated, preferably as part of a multicentre study. This, however, did not appear to be the case in the combined total of 145 patients at this centre. Furthermore, it must be borne in mind that 30% (40/132) of the group that were predicted as salvageable, died. The equation’s low sensitivity (25%) to predict mortality is desirable, as this further reduces the possibility of incorrect prediction of death. There are some very thought-provoking ethical arguments, however, that can be made in defence of such an equation. In this study, the positively predicted patients had a 100% mortality rate, despite having had DCS. In resource-limited situations, this would raise the question of whether the surgery in this subgroup of patients was futile or not. To assist in answering this question, consideration needs to be given to what there is to lose in initiating this management. Firstly, the numerous surgical and other interventions that these patients undergo, and determination of the surgical team, may give the family false hope that the patient may survive. The surgeon may be merely drawing out the inevitable and, in some instances, subjecting the patient to an undignified death. Although this is unlikely to have been an issue in most of the positively predicted patients (10/13 died within 24 h), there were a few that survived for days or even weeks. DCS also places a strain on resources such as blood products, ICU beds, theatre time, available surgeons and nursing requirements.4,16 The use of DCS on an unsalvageable patient may therefore be to the detriment of other patients, particularly in the
67
multiple-casualty setting.6 In this setting, the natural tendency of surgeons is to operate on the sickest patient first, which may be an unsalvageable patient. During this time a critically ill, but salvageable, patient may become unsalvageable if there was a delay in the damage control procedure taking place due to limited resources. In most first world settings, with abundant resources, such an equation may find little benefit. However, under-resourced hospitals commonly seen in the third world could benefit from such an objective tool, particularly when overwhelmed by extreme situations, such as natural disasters or acts of terrorism. Although this equation is an exciting prospect in the field of trauma surgery it remains only a concept at this stage and there are certain issues that need to be considered. It is based on the principle that, at some point, the patient’s physiological deterioration seems to become refractory to any form of management. However, advances in DCS techniques and post-operative resuscitation could conceivably help salvage such a patient in the future. Should such advances occur, the equation may need to be revised. Some may recommend that the equation be used retrospectively, to guide further post-operative management. The patient may, for example, undergo the initial DCS operation despite a positive prediction. If, however, the patient continues to deteriorate in the intensive care unit, the predicted outcome of the equation may help further decision making. The coefficients in this particular equation were derived from data from one centre. It may be that, with different levels of care in more advanced centres and in different population groups, the survival rates are different. The equation’s coefficients would need to be adjusted to suit that group’s particular data. It should be noted, however, that if a particular equation is derived from data from an advanced unit, which is experienced in treating DCS patients, then the potential for false positive predictions is likely to be even lower if the same equation were to be used at a less advanced unit, thereby making it safer to use. This implies that such an equation should ideally be derived from data from advanced units, preferably multiple centres, in so doing making it unlikely to give a false positive prediction in less advanced units. It is recommended that the current DCS equation described in this paper be tested in a larger, multicentre, prospective trial. This particular equation was derived in a centre with a high incidence of penetrating trauma (82%).11 Although there were no false positive predictions in patients suffering blunt trauma, testing larger cohorts of patients in this blunt trauma group would be preferable. Even if this particular equation undergoes radical change, the principle it is based on still remains, and is one that should not be discarded. That is, all human beings have a limit to what they can endure in the trauma setting, albeit with a wide variation between individuals. A point must exist however, beyond which no human being can survive their physiological decompensation and, until now we have had no way of knowing this point. The current damage control equation appears to be a step in that direction and is at the very least a preliminary definition of that point. Conclusion Despite this being one of the largest predictive studies in damage control surgery, further work and larger numbers from multiple centres are certainly required. These preliminary data, however, suggest that using the damage control equation may offer a means of predicting mortality preoperatively with a high degree of certainty. In this study, the equation predicted a quarter of all patients that died without any incorrect predictions. Although such an equation is unlikely to be used under normal circumstances, it may be useful in crisis situations when services are overwhelmed by large numbers of severely injured casualties.
68
N. Kairinos et al. / Injury, Int. J. Care Injured 41 (2010) 64–68
Conflict of interest None of the authors have any conflict of interest due to financial or personal relationships with other people or organisations that could inappropriately bias their work. There were no study sponsors. Acknowledgement Dr. Sedick Isaacs, for reviewing the statistics and assistance in formulating the equation. References 1. Aoki N, Wall MJ, Demsar J, et al. Predictive model for survival at the conclusion of a damage control laparotomy. Am J Surg 2000;180(December (6)):540–4 [discussion 544–5]. 2. Arthurs Z, Cuadrado D, Beekley A, et al. The impact of hypothermia on trauma care at the 31st combat support hospital. Am J Surg 2006;191(May (5)):610–4. 3. Asensio JA, McDuffie L, Petrone P, et al. Reliable variables in the exsanguinated patient which indicate damage control and predict outcome. Am J Surg 2001;182(December (6)):743–51. 4. Burch JM, Ortiz VB, Richardson RJ, et al. Abbreviated laparotomy and planned reoperation for critically injured patients. Ann Surg 1992;215(May (5)):476–83 [discussion 483–4]. 5. Cosgriff N, Moore EE, Sauaia A, et al. Predicting life-threatening coagulopathy in the massively transfused trauma patient: hypothermia and acidoses revisited. J Trauma 1997;42(May (5)):857–61 [discussion 861–2].
6. Eiseman B, Moore EE, Meldrum DR, Raeburn C. Feasibility of damage control surgery in the management of military combat casualties. Arch Surg 2000;135(November (11)):1323–7. 7. US Department of Defence. Emergency war surgery. 3rd ed. Virtual Naval Hospital. www.vhh.org; 2007. 8. Falcone RE, Santanello SA, Schulz MA, et al. Correlation of metabolic acidosis with outcome following injury and its value as a scoring tool. World J Surg 1993;17(September–October (5)):575–9. 9. Gregory JS, Flancbaum L, Townsend MC, et al. Incidence and timing of hypothermia in trauma patients undergoing operations. J Trauma 1991;31(June (6)):795–8 [discussion 798–800]. 10. Jurkovich GJ, Greiser WB, Luterman A, Curreri PW. Hypothermia in trauma victims: an ominous predictor of survival. J Trauma 1987;27(September (9)):1019–24. 11. Kairinos N, Nicol AJ, Timmermans J, Navsaria PH. Predicting mortality in damage control surgery. J Surg Res 2007;137(2):240. 2. 12. Krishna G, Sleigh JW, Rahman H. Physiological predictors of death in exsanguinating trauma patients undergoing conventional trauma surgery. Aust N Z J Surg 1998;68(December (12)):826–9. 13. Luna GK, Maier RV, Pavlin EG, et al. Incidence and effect of hypothermia in seriously injured patients. J Trauma 1987;27(September (9)):1014–8. 14. Morris Jr JA, Eddy VA, Blinman TA, et al. The staged celiotomy for trauma. Issues in unpacking and reconstruction. Ann Surg 1993;217(May (5)):576–84 [discussion 584–6]. 15. Rotondo MF, Schwab CW, McGonigal MD, et al. ‘Damage control’: an approach for improved survival in exsanguinating penetrating abdominal injury. J Trauma 1993;35(September (3)):375–82 [discussion 382–3]. 16. Sagraves SG, Toschlog EA, Rotondo MF. Damage control surgery—the intensivist’s role. J Intensive Care Med 2006;21(January–February (1)):5–16. 17. Shapiro MB, Jenkins DH, Schwab CW, Rotondo MF. Damage control: collective review. J Trauma 2000;49(November (5)):969–78.